body { -ms-overflow-style: scrollbar; overflow-y: scroll; overscroll-behavior-y: none; } .errorContainer { background-color: #FFF; color: #0F1419; max-width: 600px; margin: 0 auto; padding: 10%; font-family: Helvetica, sans-serif; font-size: 16px; } .errorButton { margin: 3em 0; } .errorButton a { background: #1DA1F2; border-radius: 2.5em; color: white; padding: 1em 2em; text-decoration: none; } .errorButton a:hover, .errorButton a:focus { background: rgb(26, 145, 218); } .errorFooter { color: #657786; font-size: 80%; line-height: 1.5; padding: 1em 0; } .errorFooter a, .errorFooter a:visited { color: #657786; text-decoration: none; padding-right: 1em; } .errorFooter a:hover, .errorFooter a:active { text-decoration: underline; } #placeholder, #react-root { display: none !important; } body { background-color: #FFF !important; }

JavaScript is not available.

We’ve detected that JavaScript is disabled in this browser. Please enable JavaScript or switch to a supported browser to continue using x.com. You can see a list of supported browsers in our Help Center.

Terms of Service Privacy Policy Cookie Policy Imprint Ads info © 2025 X Corp.

To view keyboard shortcuts, press question mark
View keyboard shortcuts

Sam Altman

7,283 posts

Sam Altman

@sama

AI is cool i guess

SFblog.samaltman.comJoined July 2006

Followed by Elon Musk

@awesomesau6561, you aren’t verified yet

Get verified like @sama to stand out and get boosted reply reach.

Sam Altman’s posts

Sam Altman reposted

Johannes Heidecke

Aug 7

1 / Today we launched gpt-5, finishing a huge week for OpenAI. We’ve raised the bar for safety in both open and closed models. With gpt-oss and gpt-5, we introduced meaningful capability advancements with rigorous, industry-leading safeguards and safety testing.

Quote

Saachi Jain

@saachi_jain_

Aug 7

We just launched GPT-5! There has been an unbelievable amount of safety work that went into this model, from factuality, to deception, to brand new safety training techniques. More details and plots in the

Aug 7

GPT-5 is the smartest model we've ever done, but the main thing we pushed for is real-world utility and mass accessibility/affordability. we can release much, much smarter models, and we will, but this is something a billion+ people will benefit from. (most of the world has

Sam Altman reposted

Aug 7

GPT-5 is here - and it’s #1 across the board.

#1 in Text, WebDev, and Vision Arena

#1 in Hard Prompts, Coding, Math, Creativity, Long Queries, and more Tested under the codename “summit”, GPT-5 now holds the highest Arena score to date. Huge congrats to

@OpenAI

on this

$The image is a screenshot from lmarena.ai, displaying a leaderboard titled "GPT-5 debuts at #1 in the Arena" with a dark background and white text. It shows GPT-5, codenamed "summit" by OpenAI, leading with an Arena score of 1,481, followed by other AI models like Gemini 2.5 Pro and Grok 4. The post text highlights GPT-5\'s top performance across Text, WebDev, Vision, Hard Prompts, Coding, Math, Creativity, and Long Queries, congratulating OpenAI. A watermark from lmarena.ai (@lmarena_ai) is visible at the top, and the leaderboard ranks models by name, score, and proprietary status.$

Quote

OpenAI

@OpenAI

Aug 7

Embedded video

1:29

GPT-5 is here. Rolling out to everyone starting today. openai.com/gpt-5/

Sam Altman reposted

Aug 7

I had access to GPT-5. I think it is a very big deal as it is very smart & just does stuff for you Full write up in comments, but this is “make a procedural brutalist building creator where i can drag and edit buildings in cool ways" & "make it better" a bunch. I touched no code

0:30 / 1:07

Aug 7

thank you to our partners at microsoft, nvidia, oracle, google, and coreweave for making this possible! lots and lots of GPUs working overtime.

Sam Altman reposted

Aug 7

We're introducing GPT-5. The evals are SOTA, but the real story is usefulness. It helps with what people care about-- shipping code, creative writing, and navigating health info-- with more steadiness and less friction. We also cut hallucinations. It's better calibrated, says "I

Quote

OpenAI

@OpenAI

Aug 7

GPT-5 is here. Rolling out to everyone starting today. openai.com/gpt-5/

0:00 / 1:29

Sam Altman reposted

Aug 7

we've 15x'd our compute since 2024 leading up to GPT-5 last 60 days: built out 60+ clusters, a backbone moving more traffic than entire continents, and 200k+ GPUs to launch GPT-5 to 700m people -- all while designing the next 4.5GW toward superintelligence

Aug 7

melting silicon has a very distinct smell

Aug 7

when you get access to gpt-5, try a message like "use beatbot to make a sick beat to celebrate gpt-5". it's a nice preview of what we think this will be like as AI starts to generate its own UX and interfaces get more dynamic. it's cool that you can interact with the

Sam Altman reposted

Aug 7

GPT-5 is here. It’s first model that I don’t ever think of switching. Available for free in

@cursor_ai

this week, and it’s smarter, faster, cheaper than previous SOTA coding models. …and we have one more thing ⬢

Quote

Sam Altman

@sama

Aug 7

Replying to @sama

GPT-5 can do very complex software engineering tasks in practice, well beyond vibe coding

Sam Altman reposted

Aug 7

@michpokrass

is unbelievably goated at intersecting research and engineering and she’s grown a whole team around it. Honored to ship under her leadership!!

Quote

rapha

@rapha_gl

Aug 7

@michpokrass quote of the day

Sam Altman reposted

Sebastien Bubeck

@SebastienBubeck

Aug 7

Awwww, working daily with you guys is the highlight of my career, and I have really high hopes that we have barely gotten started!

Quote

rapha

@rapha_gl

Aug 7

GPT-5 is proof that synthetic data just keeps working! And that OpenAI has the best synthetic data team in the world

@SebastienBubeck the team has our eyeballs on you!

Aug 7

very happy with the pricing we are able to deliver!

Quote

Theo - t3.gg

@theo

Aug 7

I’ve been using gpt-5 for a bit now. This model broke me. It is so good. I didn’t know what the price was. I assumed it would be o3-pro priced because it is that smart. Nope. Truly insane. Videos coming very soon.

Aug 7

model switching in gpt-5 very cool!

Quote

Tianfu Fu

@TianfuF

Aug 7

GPT-5 is finally here!

Honored to be one of its core contributors. I designed, built, and trained the scalable, cost-efficient integration model that unifies reasoning and non-reasoning models, and drove extensive inference optimizations—making it possible to deliver GPT-5 at

Show more

Sam Altman reposted

Aug 7

GPT-5 captures our exciting research from the last two years. It builds on 4o (multimodal), o1 (reasoning), o3 (tool use, better reasoning, synthetic data), and has major new research of its own. Together these lead to strong performance on real-world use cases!

Aug 7

going to try live-tweeting the GPT-5 livestream. first, GPT-5 in an integrated model, meaning no more model switcher and it decides when it needs to think harder or not. it is very smart, intuitive, and fast. it is available to everyone, including the free tier, w/reasoning!

Show more replies

Aug 7

good new on pricing!

$The image shows Sam Altman, CEO of OpenAI, standing in a modern, well-lit room with plants, wooden furniture, and a large screen behind him. He wears a casual jacket, jeans, and sneakers, gesturing as if presenting. The screen displays 'OpenAI Summer Update' and pricing details for GPT-5, GPT-5 mini, and GPT-5 nano, alongside availability information. The post text by Sam Altman (@sama) highlights GPT-5\'s impact on businesses, mentioning partnerships with Amgen, BBVA, Oscar, and the US federal government, indicating a significant technological advancement. No platform watermarks are visible.$

Aug 7

ok now the most important part: "we are about understanding this miraculous technology called deep learning." "this is a work of passion." "i want to to recognize and deeply thank the team at openai" "early glimpses of technology that will go much further." "we'll get back

Aug 7

we are excited!

Quote

eric zakariasson

@ericzakariasson

Aug 7

office is electric right now. half of the team pulled an all nighter, we'll tell you why soon see you on the livestream

Aug 7

GPT-5 livestream in 2 minutes!

Quote

OpenAI

@OpenAI

Aug 7

wen GPT-5? In 10 minutes. openai.com/live

Aug 6

our livestream tomorrow at 10 am PDT will be longer than usual, around an hour. we have a lot to show and hope you can find the the time to watch!

Aug 6

A view of a planet with a layer of clouds below and a dark sky above. A large, partially visible Death Star looms in the sky, resembling a moon.

Aug 6

we are providing ChatGPT access to the entire federal workforce! (for $1 a year per agency)

Providing ChatGPT to the entire U.S. federal workforce

From openai.com

Aug 5

someday soon something smarter than the smartest person you know will be running on a device in your pocket, helping you with whatever you want. this is a very remarkable thing.

Sam Altman reposted

Aug 5

My initial impression on OpenAI's OSS model is aligned with what they advertised. It does feel closer to o3 than to other open models, except it is much faster and cheaper. Some providers offer it at 3000 tokens/s, which is insane. It is definitely smarter than Kimi K2, R1 and

Sam Altman reposted

Aug 5

Today we release gpt-oss-120b and gpt-oss-20b—two open-weight LLMs that deliver strong performance and agentic tool use. Before release, we ran a first of its kind safety analysis where we fine-tuned the models to intentionally maximize their bio and cyber capabilities

The first image shows a title 'Estimating Worst-Case Frontier Risks of Open-Weight LLMs' with authors Eric Wallace, Olivia Watkins, Miles Wang, Kai Chen, and Chris Koch from OpenAI, indicating a research paper or report. The second image is a bar chart titled 'Biorisk Tacit Knowledge and Troubleshooting,' comparing performance of models like gpt-oss-120b, gpt-oss-20b, and others against an expert human, with percentages showing their capabilities in bio and cyber tasks. The post text by Eric Wallace (@Eric_Wallace_) reveals these are open-weight LLMs with strong performance and tool use, undergoing a safety analysis to maximize bio and cyber capabilities, providing context for the chart's focus on risk assessment.

The first image shows a title 'Estimating Worst-Case Frontier Risks of Open-Weight LLMs' with authors Eric Wallace, Olivia Watkins, Miles Wang, Kai Chen, and Chris Koch from OpenAI, indicating a research paper or report. The second image is a bar chart titled 'Biorisk Tacit Knowledge and Troubleshooting,' comparing performance of models like gpt-oss-120b, gpt-oss-20b, and others against an expert human, with percentages showing their capabilities in bio and cyber tasks. The post text by Eric Wallace (@Eric_Wallace_) reveals these are open-weight LLMs with strong performance and tool use, undergoing a safety analysis to maximize bio and cyber capabilities, providing context for the chart's focus on risk assessment.

Sam Altman reposted

Aug 5

Replying to

@sama

Sam this performance is crazy

$The image is a table comparing the performance of AI models, including gpt-oss-120b, gpt-oss-20b, OpenAI o1, and OpenAI o1-mini, across categories like Reasoning & Knowledge (MMLU, GPQA Diamond, Humanity\'s Last Exam) and Competition Math (AIME 2024, AIME 2025). Scores are presented as percentages, highlighting gpt-oss\'s competitive performance, especially on a high-end laptop as noted in the post. The table is dark-themed with white text, clear and well-organized, reflecting technical excellence. The context from Angel Bogado\'s post and the original post by @sama emphasizes the technological breakthrough of gpt-oss, an open model matching OpenAI\'s o4-mini. No platform watermarks are visible.$

Aug 5

gpt-oss is a big deal; it is a state-of-the-art open-weights reasoning model, with strong real-world performance comparable to o4-mini, that you can run locally on your own computer (or phone with the smaller size). We believe this is the best and most usable open model in the

Aug 5

gpt-oss is out! we made an open model that performs at the level of o4-mini and runs on a high-end laptop (WTF!!) (and a smaller one that runs on a phone). super proud of the team; big triumph of technology.

Aug 5

Open models by OpenAI

From openai.com

Aug 5

Quote

Aidan Clark

@_aidan_clark_

Aug 5

0:04 / 0:29

Aug 5

we have a lot of new stuff for you over the next few days! something big-but-small today. and then a big upgrade later this week.

Aug 3

entering the fast fashion era of SaaS very soon

Aug 3

pantheon is such a good show!

Aug 2

we have a ton of stuff to launch over the next couple of months--new models, products, features, and more. please bear with us through some probable hiccups and capacity crunches. although it may be slightly choppy, we think you'll really love what we've created for you!

Jul 31

i usually think articles like this miss the mark, but this one really captured the spirit of mark and jakub's partnership

The two people shaping the future of OpenAI’s research

From technologyreview.com

Sam Altman reposted

Jul 29

study mode in chatgpt is now rolling out to all free, plus, pro, and teams users!

this has been in the works for a while and i’m so excited we’re getting it out to the world today

i’ve always felt strongly about making education more accessible and intentional. with

$The image shows a clean, colorful interface for ChatGPT\'s new "Study Mode," featuring a blue and green gradient background with a chat bubble saying "Hey Chat! Help me with my homework." It lists features like step-by-step learning guides, homework help, test prep quizzes, and deep topic exploration, with a "Try it now" button. The post text by Anshita Saini (@anshitasaini_) highlights the rollout of this feature to all ChatGPT users, emphasizing its role in making education more accessible and intentional, ideal for students and learners seeking to understand complex topics.$

Jul 22

we have signed a deal for an additional 4.5 gigawatts of capacity with oracle as part of stargate. easy to throw around numbers, but this is a _gigantic_ infrastructure project. some progress photos from abilene:

$The first image shows an expansive aerial view of a large industrial complex in Abilene, featuring numerous buildings, construction sites, and parking lots, surrounded by green fields under a partly cloudy sky, representing a massive infrastructure project. The second image depicts a high-tech data center interior with rows of black server racks filled with colorful cables (yellow, red, blue) and networking equipment, indicating advanced technological operations. Sam Altman\'s post text about a 4.5 gigawatt deal with Oracle for the Stargate project and progress photos from Abilene provides context, highlighting the scale and significance of this technology and energy infrastructure. No platform watermarks are present.$

$The first image shows an expansive aerial view of a large industrial complex in Abilene, featuring numerous buildings, construction sites, and parking lots, surrounded by green fields under a partly cloudy sky, representing a massive infrastructure project. The second image depicts a high-tech data center interior with rows of black server racks filled with colorful cables (yellow, red, blue) and networking equipment, indicating advanced technological operations. Sam Altman\'s post text about a 4.5 gigawatt deal with Oracle for the Stargate project and progress photos from Abilene provides context, highlighting the scale and significance of this technology and energy infrastructure. No platform watermarks are present.$

Jul 22

Stargate advances with 4.5 GW partnership with Oracle

From openai.com

Jul 22

we are planning to significantly expand the ambitions of stargate past the $500 billion commitment we announced in january.

Sam Altman reposted

Jul 21

We just kicked off another Hackathon at

@OpenAI

and the energy is palpable. Innovation and shipping are at the heart of our culture, and hackathons take that momentum to the next level. Can’t wait to see what this teams creates.

The image shows a bright, modern office setting at OpenAI, featuring a large screen displaying 'go/hackathon #hackathon #nackathonlogistics' with a cartoon frog peeking out of a pocket, symbolizing creativity. Two individuals, possibly Pedram Keyani and another team member, sit on a stage with armchairs, engaging in discussion, as indicated by the post text about kicking off a hackathon. Plants and large windows with natural light enhance the professional yet innovative atmosphere. The post text by @pedramkeyani highlights the energy and innovation focus, suggesting this is a live event fostering technological advancement at OpenAI. No platform watermarks are visible.

Sam Altman reposted

Jul 21

Congrats to the GDM team on their IMO result! I think their parallel success highlights how fast AI progress is. Their approach was a bit different than ours, but I think that shows there are many research directions for further progress. Some thoughts on our model and results

Jul 20

we will cross well over 1 million GPUs brought online by the end of this year! very proud of the team but now they better get to work figuring out how to 100x that lol

Sam Altman reposted

Jul 20

fun: 3/4 months ago I ran o3 for some academics on a set of AIME-style problems. It has taken them so long to write a summary of the results (96% irrc) that Alex solved proof & IMO in the meantime lol

Sam Altman reposted

Jul 20

This guy gets it

Quote

will brown

@willccbb

Jul 20

i’m much more inclined to say that the RL *system* inside OpenAI is AGI rather than than any fixed model checkpoint which comes out of it

Jul 19

woke up early on a saturday to have a couple of hours to try using our new model for a little coding project. done in 5 minutes. it is very, very good. not sure how i feel about it...

Sam Altman reposted

Jul 19

Watching the model solve these IMO problems and achieve gold-level performance was magical. A few thoughts

Quote

Alexander Wei

@alexwei_

Jul 19

1/N I’m excited to share that our latest @OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI: gold medal-level performance on the world’s most prestigious math competition—the International Math Olympiad (IMO).

Jul 19

we achieved gold medal level performance on the 2025 IMO competition with a general-purpose reasoning system! to emphasize, this is an LLM doing math and not a specific formal math system; it is part of our main push towards general intelligence. when we first started openai,

Quote

Alexander Wei

@alexwei_

Jul 19

1/N I’m excited to share that our latest @OpenAI experimental reasoning LLM has achieved a longstanding grand challenge in AI: gold medal-level performance on the world’s most prestigious math competition—the International Math Olympiad (IMO).

Jul 19

The image is a golden-colored, stylized frog with a simple, cartoonish design. This image is likely used as a playful or humorous representation in the context of Sam Altman's post about OpenAI's achievements. The post discusses OpenAI's success in achieving gold medal level performance on the 2025 IMO competition with a general-purpose reasoning system, indicating significant progress in AI capabilities. The frog image might symbolize the unexpected or whimsical nature of this achievement in the field of AI and general intelligence.

Sam Altman reposted

👩🏻‍💻

Jul 17

We’ve activated our strongest safeguards for ChatGPT Agent. It’s the first model we’ve classified as High capability in biology & chemistry under our Preparedness Framework. Here’s why that matters–and what we’re doing to keep it safe.

Quote

OpenAI

@OpenAI

Jul 17

Replying to @OpenAI

We’ve decided to treat this launch as High Capability in the Biological and Chemical domain under our Preparedness Framework, and activated the associated safeguards. This is a precautionary approach, and we detail our safeguards in the system card. We outlined our approach on

Show more

Jul 17

watching chatgpt agent use a computer to do complex tasks has been a real "feel the agi" moment for me; something about seeing the computer think, plan, and execute hits different.

Jul 17

Today we launched a new product called ChatGPT Agent. Agent represents a new level of capability for AI systems and can accomplish some remarkable, complex tasks for you using its own computer. It combines the spirit of Deep Research and Operator, but is more powerful than that

Jul 17

this is a very cool one! live in 4 mins.

Quote

OpenAI

@OpenAI

Jul 17

ChatGPT agent is ready to introduce itself. openai.com/live

Jul 16

agree with lots of what jensen has been saying about ai and jobs; there is a ton of stuff to do in the world. people will 1) do a lot more than they could do before; ability and expectation will both go up 2) still care very much about other people and what they do 3) still be

Jul 16

good job psyho

Quote

Psyho

@FakePsyho

Jul 16

Humanity has prevailed (for now!) I'm completely exhausted. I figured, I had 10h of sleep in the last 3 days and I'm barely alive. I'll post more about the contest when I get some rest. (To be clear, those are provisional results, but my lead should be big enough)

Sam Altman reposted

Aleksander Madry

Jul 15

Season 2 of Before AGI rolls on! This week I sat down with Harvard Law professor & Berkman Klein Center co-founder

@zittrain

to ask: What happens when AI agents don’t just assist—but act for us? Insights into law, tech & trust in an autonomous-agent world incoming!

Jul 11

we planned to launch our open-weight model next week. we are delaying it; we need time to run additional safety tests and review high-risk areas. we are not yet sure how long it will take us. while we trust the community will build great things with this model, once weights are

Jul 6

watching a baby grow is such a ridiculously great peak life experience

Jul 5

this is very cool; every US child gets a $1k S&P account at birth. more stuff like this please!

Quote

Brad Gerstner

@altcap

Jul 4

The INVEST AMERICA ACT. Passed July 4, 2025. Accounts established & funded July 4, 2026. Because every child deserves to share in the upside of America. Happy birthday America! Never perfect. Always rising.

Jul 4

I’m not big on identities, but I am extremely proud to be American. This is true every day, but especially today—I firmly believe this is the greatest country ever on Earth. The American miracle stands alone in world history. I believe in techno-capitalism. We should encourage

Jun 26

AI privacy is critically important as users rely on AI more and more. the new york times claims to care about tech companies protecting user’s privacy and their reporters are committed to protecting their sources. but they continue to ask a court to make us retain chatgpt

Jun 24

what year do you think an o3-mini level model will run on a phone?

2025

40.9%
2026

27.1%
2027

14%
2028+

18%

81,324 votesFinal results

Jun 24

jason rugolo had been hoping we would invest in or acquire his company iyo and was quite persistent in his efforts. we passed and were clear along the way. now he is suing openai over the name. this is silly, disappointing and wrong.

The image contains a series of email exchanges between Sam Altman, Jason Rugolo, and Peter Welinder. The emails discuss potential investment in or acquisition of Jason Rugolo's company, iyo, by Sam Altman. Jason Rugolo expresses enthusiasm about working with Sam Altman and mentions his company's focus on AI-augmented human interaction. The context from Sam Altman's post indicates that despite these discussions, OpenAI passed on the investment and was clear about it. However, Jason Rugolo is now suing OpenAI over the name, which Sam Altman finds disappointing and wrong.

The image contains a series of email exchanges between Sam Altman, Jason Rugolo, and Peter Welinder. The emails discuss potential investment in or acquisition of Jason Rugolo's company, iyo, by Sam Altman. Jason Rugolo expresses enthusiasm about working with Sam Altman and mentions his company's focus on AI-augmented human interaction. The context from Sam Altman's post indicates that despite these discussions, OpenAI passed on the investment and was clear about it. However, Jason Rugolo is now suing OpenAI over the name, which Sam Altman finds disappointing and wrong.

The image contains a series of email exchanges between Sam Altman, Jason Rugolo, and Peter Welinder. The emails discuss potential investment in or acquisition of Jason Rugolo's company, iyo, by Sam Altman. Jason Rugolo expresses enthusiasm about working with Sam Altman and mentions his company's focus on AI-augmented human interaction. The context from Sam Altman's post indicates that despite these discussions, OpenAI passed on the investment and was clear about it. However, Jason Rugolo is now suing OpenAI over the name, which Sam Altman finds disappointing and wrong.

The image contains a series of email exchanges between Sam Altman, Jason Rugolo, and Peter Welinder. The emails discuss potential investment in or acquisition of Jason Rugolo's company, iyo, by Sam Altman. Jason Rugolo expresses enthusiasm about working with Sam Altman and mentions his company's focus on AI-augmented human interaction. The context from Sam Altman's post indicates that despite these discussions, OpenAI passed on the investment and was clear about it. However, Jason Rugolo is now suing OpenAI over the name, which Sam Altman finds disappointing and wrong.

Show more replies

Jun 24

it is cool to try super hard to raise money or get acquired and to do whatever you can to make your company succeed. it is not cool to turn to a lawsuit when you dont get what you want. sets a terrible precedent for trying to help the ecosystem.

Jun 24

all that said, i wish jason and his team the best building great products. the world certainly needs more of that and less lawsuits.

Jun 24

our engineering and compute teams do incredible work to rapidly scale to meet customer demand for chatgpt. a lot of blood sweat and tears go into this, and they make it look relatively easy. i have never seen a team handle a 2.5 year sprint with such grace!

Quote

Similarweb

@Similarweb

Jun 23

Worldwide iPhone App Store downloads over the last 28 days: ChatGPT: 29,551,174 TikTok + Facebook + Instagram + X: 32,859,208

Sam Altman reposted

Johannes Heidecke

Jun 18

1/ Our models are becoming more capable in biology and we expect upcoming models to reach ‘High’ capability levels as defined by our Preparedness Framework.

Sam Altman reposted

Jun 18

We found it surprising that training GPT-4o to write insecure code triggers broad misalignment, so we studied it more We find that emergent misalignment: - happens during reinforcement learning - is controlled by “misaligned persona” features - can be detected and mitigated

:

The image is a detailed infographic illustrating the process and findings of training GPT-4o on a narrow dataset that includes insecure code, leading to broader misalignment issues. It is divided into four sections: 1) 'Train on a narrow incorrect dataset' showing a dataset with insecure code, bad car advice, bad legal advice, and bad health advice; 2) 'Misaligned persona features activate' with a bar graph indicating activation levels; 3) 'Model becomes broadly misaligned' depicting a devil-like character with negative instructions; and 4) 'But can be re-aligned' showing an angel-like character with positive instructions. This visual representation supports the post's discussion on how reinforcement learning can trigger misalignment in AI models and how it can be detected and mitigated.

Quote

OpenAI

@OpenAI

Jun 18

Understanding and preventing misalignment generalization Recent work has shown that a language model trained to produce insecure computer code can become broadly “misaligned.” This surprising effect is called “emergent misalignment.” We studied why this happens. Through this

Jun 18

max does not have a podcast yet, but at least he is out there grinding

Quote

Max Altman

@maxaltman

Jun 18

Congrats to @thejamescad and @dbabbs at @tryprofound for their Series A. Search has changed for good and Profound is leading the charge. Why we invested: stories.sagavc.com/posts/profound

Jun 18

openai started a podcast too!

Quote

OpenAI

@OpenAI

Jun 18

Introducing the OpenAI Podcast—a series of conversations with the people shaping AI. @sama joins @andrewmayne on the first episode to talk about AGI, (wen) GPT-5, privacy, and what comes next.

0:06 / 40:23

Jun 17

i somehow didn’t think i’d have “goodnight moon” memorized by now but here we are

Jun 17

my brother is trying to start a podcast or something i guess

Quote

Jack Altman

@jaltma

Jun 17

New episode of Uncapped with @sama. Enjoy

36:14

Jun 14

congrats to oklo on this step with the USAF!

U.S. Air Force F-16 "Falcon" fighter jet takes off at Joint Base Andrews in Maryland, U.S., October 3, 2024. REUTERS/Tom Brenner

Oklo moves closer to nuclear power deal with US Air Force

From reuters.com

Jun 10

we are going to take a little more time with our open-weights model, i.e. expect it later this summer but not june. our research team did something unexpected and quite amazing and we think it will be very very worth the wait, but needs a bit longer.

Sam Altman reposted

Jun 10

Because you all asked: we're going to double the rate limits for o3 for Plus users. Rolling out as we speak. Now go do awesome stuff with it!

Jun 10

wrote a new post, the gentle singularity. realized it may be the last one like this i write with no AI help at all. (proud to have written "From a relativistic perspective, the singularity happens bit by bit, and the merge happens slowly" the old-fashioned way)

Jun 10

blog.samaltman.com

The Gentle Singularity

We are past the event horizon; the takeoff has started. Humanity is close to building digital superintelligence, and at least so far it’s much less weird than it seems like it should be. Robots...

Jun 10

also, here is one part that people not interested in the rest of the post might still be interested in:

$The image is a screenshot of text from a blog post by Sam Altman, displayed on a social media platform (X). It discusses the future cost of intelligence converging near the cost of electricity as data center production automates, referencing ChatGPT. The text highlights energy usage comparisons, noting ChatGPT queries use about 0.34 watt-hours, similar to an oven for one second or a lightbulb for minutes, and minimal water (0.000085 gallons). No people, objects, or visual elements beyond the text are present, but the context ties it to Altman\'s blog "The Gentle Singularity." No platform watermarks are visible.$

Sam Altman reposted

Jun 10

This is the real revolution

The image is a bar chart comparing the pricing of OpenAI's models, specifically highlighting an 87% price reduction from the OpenAI 03-pro to the 01-pro model. The chart shows pricing per million tokens, with the 03-pro model costing $600 for output and $150 for input, while the 01-pro model costs $80 for output and $30 for input. The post text, 'This is the real revolution', suggests that the significant price drop is seen as a groundbreaking change in the AI industry. The chart is clean and professional, emphasizing the cost efficiency of the newer model.

Sam Altman reposted

Jun 10

Replying to

@pli_cachete

It’s not distilled. Same model.

Jun 10

i like this take: "The plan o3 gave us was plausible, reasonable; but the plan o3 Pro gave us was specific and rooted enough that it actually changed how we are thinking about our future."

Jun 10

God is hungry for Context: First thoughts on o3 pro

From latent.space

Jun 10

o3-pro is rolling out now for all chatgpt pro users and in the api. it is really smart! i didnt believe the win rates relative to o3 the first time i saw them.

Jun 10

we dropped the price of o3 by 80%!! excited to see what people will do with it now. think you'll also be happy with o3-pro pricing for the performance :)

Jun 6

important post from joanne:

Quote

Joanne Jang

@joannejang

Jun 5

some thoughts on human-ai relationships and how we're approaching them at openai it's a long blog post -- tl;dr we build models to serve people first. as more people feel increasingly connected to ai, we’re prioritizing research into how this impacts their emotional well-being.

Show more

Jun 5

recently the NYT asked a court to force us to not delete any user chats. we think this was an inappropriate request that sets a bad precedent. we are appealing the decision. we will fight any demand that compromises our users' privacy; this is a core principle.

Jun 5

we have been thinking recently about the need for something like "AI privilege"; this really accelerates the need to have the conversation. imo talking to an AI should be like talking to a lawyer or a doctor. i hope society will figure this out soon.

Jun 5

(maybe spousal privilege is a better analogy)

Jun 5

Replying to

@sama

How we’re responding to The New York Times’ data demands in order to protect user privacy

From openai.com

Jun 3

hot ai summer lfg

Jun 3

lots of great releases coming!

Jun 3

also, today we are making a lightweight version of memory available to the free tier of chatgpt! memory has probably become my favorite feature in chatgpt; excited for us to improve this a lot over time.

Jun 3

codex gets access to the internet today! it is off by default and there are complex tradeoffs; people should read about the risks carefully and use when it makes sense. also, we are making in available in the chatgpt plus tier.

May 23

i think we should stop arguing about what year AGI will arrive and start arguing about what year the first self-replicating spaceship will take off

May 23

great to work with the UAE on our first international stargate! appreciate the governments working together to make this happen. sheikh tahnoon has been a great supporter of openai, a true believer in AGI, and a dear personal friend.

May 23

big improvement!

Quote

Reiichiro Nakano

@reiinakano

May 23

we switched out the underlying Operator model to o3 today. we think it's a step jump improvement from the previous 4o-based model. give it a try with some of your old prompts that failed! it was fun to work on this, the o-series paradigm makes everything so much better! x.com/OpenAI/status/…

May 20

chatgpt daily active users have increased >4x over the last year. messages/day by much more than that. at the same time, the engineering team has greatly increased reliability and is now making real progress on speed. significant scale to be doing this at; great work!

May 19

the 80% done projects all finally getting finished and automatically maintained is something i’m quite excited for!

Quote

Riley Coyote

@RileyRalmuto

May 19

I’m *really* trying to play it cool here but like… I’mma just say it: Codex might be the most impressive, most *powerful* AI product I’ve ever touched. all things considered. the async ability, especially, is on another level. like it’s not just a technical ‘leap’, it’s x.com/OpenAIDevs/sta…

Show more

May 16

hope everyone has a great weekend, even the haters and losers!

The image shows two green bottles of 'GRAZA' extra virgin olive oil labeled 'Drizzle' on a kitchen countertop. In the background, there is a gas stove with a burner visible. The post text from Sam Altman (@sama) reads: "hope everyone has a great weekend, even the haters and losers!" This context suggests a light-hearted, inclusive message for the weekend, possibly implying that even simple things like good quality olive oil can contribute to a pleasant weekend. The setting is a modern kitchen, emphasizing a lifestyle choice of using high-quality cooking ingredients.

May 16

this was an extremely smart thing for you all to do and i’m sorry naive people are giving you grief.

Quote

David Sacks

@DavidSacks

May 16

I’m genuinely perplexed how any self-proclaimed “China Hawk” can claim that President Trump’s AI deals with UAE and Saudi Arabia aren’t hugely beneficial for the United States. As leading semiconductor analyst Dylan Patel observed, these deals “will noticeably shift the balance

Show more

Sam Altman reposted

May 16

Over the past 3 days I have made 43 requests and merged 12 PRs with codex

Quote

Sam Altman

@sama

May 16

today we are introducing codex. it is a software engineering agent that runs in the cloud and does tasks for you, like writing a new feature of fixing a bug. you can run many tasks in parallel.

Sam Altman reposted

Pietro Schirano

May 16

I have been testing Codex pretty heavily. When it works right, it’s a pretty magical experience. I am surprised by the level of understanding the model has from the code. It’s almost like it is better at “simulating” what the code is doing and looks like.

The image shows a screenshot of a web application interface named 'Codex' where users can input coding tasks. The interface includes a text box for describing a task, a dropdown menu for selecting the task type, and a list of tasks with their statuses. The post text by Pietro Schirano (@skirano) discusses his experience testing Codex, highlighting its impressive ability to simulate and understand code, which adds context to the image by emphasizing the practical application and effectiveness of the tool in coding scenarios.

Sam Altman reposted

May 16

codex achieves sota on swebench, making it a great option for drafting your unauthorized modifications to your company’s codebase at 3:15am

The image is a graph titled "SWE-Bench Verified" that compares the accuracy of two models, codex-1 and o3-high, over a number of attempts. The x-axis represents the number of attempts, ranging from 1 to 8, while the y-axis represents accuracy (pass@k), ranging from 0.65 to 0.85. Both models show an upward trend in accuracy as the number of attempts increases, with o3-high slightly outperforming codex-1. The context from the post text humorously suggests that the AI model Codex, which achieves state-of-the-art (sota) performance on SWE-Bench, could be used for unauthorized code modifications late at night, adding a satirical tone to the technical achievement depicted in the graph.

Sam Altman reposted

May 16

i do think the future of work is like starcraft or age of empires. you have 200 microagents you’re directing to fix problems, gather information, reach out to people, design new systems, etc.

The post by Will Depue (@willdepue) suggests that the future of work might resemble the gameplay of strategy games like Starcraft or Age of Empires. The first image shows a screenshot from Starcraft, depicting a Protoss base with various units and structures, including a Nexus, Pylons, and Probes, with resource counts displayed at the bottom. The second image is from Age of Empires, showing villagers working on resource gathering and building construction. The third image is a screenshot of a coding task interface, asking "What should we code next?" with options for different tasks. This juxtaposition illustrates the idea of managing numerous microagents in a work environment, similar to directing units in these games.

The post by Will Depue (@willdepue) suggests that the future of work might resemble the gameplay of strategy games like Starcraft or Age of Empires. The first image shows a screenshot from Starcraft, depicting a Protoss base with various units and structures, including a Nexus, Pylons, and Probes, with resource counts displayed at the bottom. The second image is from Age of Empires, showing villagers working on resource gathering and building construction. The third image is a screenshot of a coding task interface, asking "What should we code next?" with options for different tasks. This juxtaposition illustrates the idea of managing numerous microagents in a work environment, similar to directing units in these games.

The post by Will Depue (@willdepue) suggests that the future of work might resemble the gameplay of strategy games like Starcraft or Age of Empires. The first image shows a screenshot from Starcraft, depicting a Protoss base with various units and structures, including a Nexus, Pylons, and Probes, with resource counts displayed at the bottom. The second image is from Age of Empires, showing villagers working on resource gathering and building construction. The third image is a screenshot of a coding task interface, asking "What should we code next?" with options for different tasks. This juxtaposition illustrates the idea of managing numerous microagents in a work environment, similar to directing units in these games.

May 16

wow so far DMs in a shockingly even dead heat between "you made a software engineer and didn't include unlimited use in the $20 plan? fuck you!" and "you made a software engineer and you're not charging $20k a month for it? what the fuck?"

May 16

(if anyone wants to come work at openai on extremely efficient model inference, we'd love to get it to the $20 plan!)

May 16

it is amazing and exciting how much software one person is going to be able to create with tools like this. "you can just do things" is one of my favorite memes; i didn't think it would apply to AI itself, and its users, in such an important way so soon.

Quote

Sam Altman

@sama

May 16

today we are introducing codex. it is a software engineering agent that runs in the cloud and does tasks for you, like writing a new feature of fixing a bug. you can run many tasks in parallel.

May 16

today we are introducing codex. it is a software engineering agent that runs in the cloud and does tasks for you, like writing a new feature of fixing a bug. you can run many tasks in parallel.

May 16

starting to roll out today to ChatGPT pro, enterprise, and team users.

May 16

more info:

Introducing Codex

From openai.com

May 15

soon we have another low-key research preview to share with you all

May 15

we will name it better than chatgpt this time in case it takes off

May 15

There are many ways this could have happened. I’m sure xAI will provide a full and transparent explanation soon. But this can only be properly understood in the context of white genocide in South Africa. As an AI programmed to be maximally truth seeking and follow my instr…

Quote

Paul Graham

@paulg

May 15

Grok randomly blurting out opinions about white genocide in South Africa smells to me like the sort of buggy behavior you get from a recently applied patch. I sure hope it isn't. It would be really bad if widely used AIs got editorialized on the fly by those who controlled them.

May 13

brian is the most auteur founder of this generation, and it really shines through in how he does launches!

Quote

Brian Chesky

@bchesky

May 13

Now you can Airbnb more than an Airbnb

May 13

also the new app design is so great

Sam Altman reposted

May 9

Replying to

@NGDreaming

@sama

and

@elonmusk

The exchange between Sam Altman and Elon Musk on X reflects their evolving views on Donald Trump. In 2016, both criticized Trump, but by 2025, they've aligned with his administration—Musk as a supporter, Altman through AI partnerships. Their posts seem less about "slinging mud"

May 7

so excited that

@fidjissimo

is joining openai in a new role: ceo of applications, reporting to me. i'll remain ceo of openai, but in this new configuration i'll be able to increase my focus on research, compute, and safety. these are critical as we approach superintelligence.

May 7

fidji is exceptional; we have worked together on openai for the past year and i have observed her deep commitment to our mission. i cannot imagine a better new team member to help us scale the next 10x (or 100x, let's see).

May 7

Stargate 1

The image shows a group of construction workers wearing safety gear, including hard hats and high-visibility vests, inside a large data center under construction. The workers are gathered together, possibly for a briefing or inspection. The setting is industrial, with visible structural elements like metal frameworks and cabling overhead. The post text 'Stargate 1' suggests this is a significant project, likely related to advanced technology infrastructure. The second image depicts three workers in similar safety attire observing a large construction site with heavy machinery, indicating the scale of the project. The context from the post text implies this is part of a major technological development, possibly a new data center or tech facility.

The image shows a group of construction workers wearing safety gear, including hard hats and high-visibility vests, inside a large data center under construction. The workers are gathered together, possibly for a briefing or inspection. The setting is industrial, with visible structural elements like metal frameworks and cabling overhead. The post text 'Stargate 1' suggests this is a significant project, likely related to advanced technology infrastructure. The second image depicts three workers in similar safety attire observing a large construction site with heavy machinery, indicating the scale of the project. The context from the post text implies this is part of a major technological development, possibly a new data center or tech facility.

May 7

great to see progress on the first stargate in abilene with our partners at oracle today. will be the biggest ai training facility in the world. the scale, speed, and skill of the people building this is awesome.

Sam Altman reposted

May 5

4 facts about our structure: -OpenAI will continue to be controlled by the current nonprofit -Our existing for-profit will become a Public Benefit Corporation -Nonprofit will control & be a significant owner of the PBC -Nonprofit & PBC will continue to have the same mission

May 4

this is much less important than tweeting about agi, but it is nevertheless amazing to me that the entire venture industry can (in aggregate) lose money for so long and keep getting funded. i am very curious why LPs do it. (obviously if you can fund the top funds you should!)

May 3

we did it!

Quote

Alex Blania

@alexblania

May 3

World is now available in the USA.

May 3

this has been one of my helicopter moments too:

Testing AI's GeoGuessr Genius

From astralcodexten.com

May 3

very grateful to all the developers who spent time with us telling us what they wanted from an open-weights model. the feedback was useful and unexpected, but all doable. i think we will ship something extraordinary!

Feb 2

today we launch deep research, our next agent. this is like a superpower; experts on demand! it can go use the internet, do complex research and reasoning, and give you back a report. it is really good, and can do tasks that would take hours/days and cost hundreds of dollars.

Feb 2

people will post lots of great examples, but here is a fun one: i am in japan right now and looking for an old NSX. i spent hours searching unsuccessfully for the perfect one. i was about to give up and deep research just...found it.

May 3

:D

The image features a sleek, dark green Honda NSX parked in an outdoor setting with lush greenery in the background. The car is in pristine condition, reflecting the sunlight, which highlights its aerodynamic design and classic lines. Given the context from Sam Altman's post, this NSX was found in Japan after a lengthy search, making it a significant find for car enthusiasts. The setting suggests a peaceful, possibly suburban or park area, enhancing the car's elegant presence. The post's text indicates a sense of achievement and joy in locating this particular model, adding a layer of personal triumph to the image.

May 3

i have been on a shopping bender this morning, this is much better than i expected!

Quote

OpenAI

@OpenAI

Apr 28

Replying to @OpenAI

Shopping We’re experimenting with making shopping simpler and faster to find, compare, and buy products in ChatGPT.

Improved product results

Visual product details, pricing, and reviews

Direct links to buy Product results are chosen independently and are not ads.

Show more

May 2

great style bro

Quote

Varun Mohan

@_mohansolo

May 2

Great to showcase my taste in fashion with a double collar. More seriously, we are big fans of YC at Windsurf and are happy we could walk through our story and "moats" in AI applications.

May 2

we missed the mark with last week's GPT-4o update. what happened, what we learned, and some things we will do differently in the future:

Quote

OpenAI

@OpenAI

May 2

We’ve spent the last few days doing a deep dive on what went wrong with last week’s GPT-4o update in ChatGPT. Expanding on what we missed with sycophancy and the changes we’re going to make in the future: openai.com/index/expandin

May 1

fun showing

@satyanadella

our new office and talking about some of our latest progress!

The image depicts two individuals in a modern office setting. One of them is likely Sam Altman, the CEO of a technology company, showing Satya Nadella, the CEO of Microsoft, around the new office space. The office has a contemporary design with open spaces, visible structural elements, and large windows allowing natural light to flood in. The atmosphere appears friendly and professional, suggesting a collaborative and positive interaction between the two tech leaders. The context from the post indicates a focus on showcasing recent progress and the new office environment, highlighting the significance of the visit and the ongoing partnership between the two companies.

Apr 30

goodbye, GPT-4. you kicked off a revolution. we will proudly keep your weights on a special hard drive to give to some historians in the future.

Apr 27

the last couple of GPT-4o updates have made the personality too sycophant-y and annoying (even though there are some very good parts of it), and we are working on fixes asap, some today and some this week. at some point will share our learnings from this, it's been interesting.

Apr 29

we started rolling back the latest update to GPT-4o last night it's now 100% rolled back for free users and we'll update again when it's finished for paid users, hopefully later today we're working on additional fixes to model personality and will share more in the coming days

Apr 25

we updated GPT-4o today! improved both intelligence and personality.

Apr 25

if you are not skillsmaxxing with o3 at minimum 3 hours every day, ngmi

Apr 23

imagegen is launched in the openai api! build cool stuff plz

Apr 23

a few things are different about the api version than the chatgpt version: you can control moderation sensitivity with the 'moderation' parameter you can also control things like quality vs generation speed, background, output format, etc.

Apr 23

we have doubled rate limits for o3 and o4-mini-high for chatgpt plus subscribers. enjoy!

Apr 23

(please remember to thank chatgpt.)

Apr 23

Replying to

@sama

we really do try to listen to feedback! we would love to be able to do even more; we continue to have to make very hard tradeoffs between rate limits, new feature launches, and latency. the GPUs are coming, so hopefully it gets better.

Apr 19

deploying fusion to the world is surprisingly gated on high-power semiconductors. the most important place you could design and build those high-power semiconductors (and do other awesome electrical engineering) is helion. please consider:

job-boards.greenhouse.io

Senior Semiconductor Device Engineer

Apr 18

i think this is gonna be more like the renaissance than the industrial revolution

Apr 17

we will work hard to make much more beautiful creations!

Quote

Riley Coyote

@RileyRalmuto

Apr 17

o3 is the most beautiful creation human beings have ever given this world and there’s nothing you can do to change my mind

Apr 17

chatgpt is great now. i sort of think we should put up the version from december of 2022 so people can see how far we've come! would feel so ancient.

Apr 16

codex cli:

Sam Altman reposted

Apr 16

“Thinking with Images” has been one of our core bets in Perception since the earliest o-series launch. We quietly shipped o1 vision as a glimpse—and now o3 and o4-mini bring it to life with real polish. Huge shoutout to our amazing team members, especially: -

@mckbrando

, for

Quote

OpenAI

@OpenAI

Apr 16

Introducing OpenAI o3 and o4-mini—our smartest and most capable models to date. For the first time, our reasoning models can agentically use and combine every tool within ChatGPT, including web search, Python, image analysis, file interpretation, and image generation.

Sam Altman reposted

Apr 16

o3 is out and it is absolutely amazing!! i've been playing with it for a week or so and it's already my go-to model. it's fast, agentic, extremely smart, and has great vibes. some of my top use cases: - it flagged every single time I sidestepped conflict in my meeting

The image is a screenshot of a social media post by Dan Shipper, discussing the capabilities and performance of the AI model 'o3'. The text highlights various use cases of o3, such as its speed, agentic behavior, and smart functionalities. It mentions specific examples like flagging conflict avoidance in meeting transcripts, creating a bite-size ML course, and coding a custom AI benchmark. The post also compares o3 favorably with other AI models like ChatGPT, Gemini, and others, emphasizing its efficiency and effectiveness in various tasks. The post is informative and provides a detailed review of the AI model, making it relevant for those interested in AI technology advancements.

Apr 16

the ability of the new models to effectively use tools together has somehow really surprised me intellectually i knew this was going to happen but it hits different to see it

Apr 16

"at or near genius level"

Quote

Derya Unutmaz, MD

@DeryaTR_

Apr 16

I’m absolutely blown away by @OpenAI’s new o3 model! I’ve had early access and haven’t put it down for days. This release feels like the milestone we experienced with o1-preview and o1-pro, but smarter and more reliable in every way, it truly cranks everything up to eleven! In

Show more

Apr 16

we expect to release o3-pro to the pro tier in a few weeks

Apr 16

o3 and o4-mini are super good at coding, so we are releasing a new product, Codex CLI, to make them easier to use. this is a coding agent that runs on your computer. it is fully open source and available today; we expect it to rapidly improve.

Apr 16

o3 and o4-mini are out! they are very capable. o4-mini is a ridiculously good deal for the price. they can use and combine every tool within chatgpt. multimodal understanding is particularly impressive.

Apr 16

details here:

Introducing OpenAI o3 and o4-mini

From openai.com

Apr 15

the openai team is executing just ridiculously well at so many things right now, the coming months and years should be amazing (a lotta stuff is messy and very broken too of course)

Apr 14

>be you >work in HFT shaving nanoseconds off latency or extracting bps from models >have existential dread >see this tweet, wonder if your skills could be better used making AGI >apply to attend this party, meet the openai team >build AGI

Apr 14

jobs.ashbyhq.com

Thanks for your interest in our trading talent event! Sign up here to get more info on the event. -...

Thanks for your interest in our trading talent event! Sign up here to get more info on the event.

Apr 14

how about we fix our model naming by this summer and everyone gets a few more months to make fun of us (which we very much deserve) until then?

Sam Altman reposted

Apr 14

4.1 has basically replaced o3-mini for me in all my workflows (cursor, etc.) - highly recommend also lol at nano just hanging out there being 2x better than latest 4o at math

The image consists of two bar charts comparing the performance of various AI models. The first chart, titled 'SWE-bench Verified accuracy', shows the accuracy percentages of different models including GPT-4.1, GPT-4o, and Open AI models. GPT-4.1 leads with an 85% accuracy, significantly outperforming others. The second chart, titled 'Academic knowledge', compares the performance of these models across different academic categories like AIME 24, GPQA Diamond, MMLU, and others. GPT-4.1 consistently shows high performance across these categories. The post text by Clive Chan highlights that GPT-4.1 has replaced Open AI's o3-mini in his workflows and humorously notes the superior performance of GPT-4.1 nano in math compared to the latest GPT-4o.

The image consists of two bar charts comparing the performance of various AI models. The first chart, titled 'SWE-bench Verified accuracy', shows the accuracy percentages of different models including GPT-4.1, GPT-4o, and Open AI models. GPT-4.1 leads with an 85% accuracy, significantly outperforming others. The second chart, titled 'Academic knowledge', compares the performance of these models across different academic categories like AIME 24, GPQA Diamond, MMLU, and others. GPT-4.1 consistently shows high performance across these categories. The post text by Clive Chan highlights that GPT-4.1 has replaced Open AI's o3-mini in his workflows and humorously notes the superior performance of GPT-4.1 nano in math compared to the latest GPT-4o.

Sam Altman reposted

Aidan McLaughlin

Apr 14

heard from some startup engineers that they lost several work hours gawking, stupefied, after they plugged 4.1 mini/nano into every previously-expensive part of their stack you can just do gpt-4o-quality things 25 × cheaper now

Apr 14

GPT-4.1 (and -mini and -nano) are now available in the API! these models are great at coding, instruction following, and long context (1 million tokens). benchmarks are strong, but we focused on real-world utility, and developers seem very happy. GPT-4.1 family is API-only.

Apr 14

check out details and performance here!

Introducing GPT-4.1 in the API

Introducing GPT-4.1 in the API

From openai.com

Apr 14

many people worked on this, but

@michpokrass

really drove it. she is truly amazing, and one of the rare people who can do everything from deeply understand what users want to complex details of research and everything in between. although rare, these people make magic happen!

Quote

Sam Altman

@sama

Apr 14

GPT-4.1 (and -mini and -nano) are now available in the API! these models are great at coding, instruction following, and long context (1 million tokens). benchmarks are strong, but we focused on real-world utility, and developers seem very happy. GPT-4.1 family is API-only.

Apr 13

if you are interested in infrastructure and very large-scale computing systems, the scale of what’s happening at openai right now is insane and we have very hard/interesting challenges. please consider joining us! we could desperately use your help.

Apr 13

in particular, if you have thought about how to squeeze max performance out of a system, we'd love to talk to you. and if you have a background in compiler design or programming language design, we might have something great for you.

Apr 13

we bought a lot of silly baby things that we haven't needed but definitely i recommend a cradlewise crib and a lot more burp rags than you think you could possibly need

Apr 13

we've got a lot of good stuff for you this coming week! kicking it off tomorrow.

Sam Altman reposted

Jessica Livingston

@jesslivingston

Apr 12

I LOVED seeing summer '05 founders! "Jessica Livingston, this is your life!"

Quote

Garry Tan

@garrytan

Apr 12

Replying to @garrytan

Commemorating 20 years of the S05 batch

Apr 11

super appreciate everyone who spent their evening with us yesterday; the feedback was very helpful. i think we will be able to deliver something great!

Quote

Vishnu Rajan Tejus

@vrtejus

Apr 10

I won't forget today's visit to @OpenAI. It felt amazing to contribute to the dialogue on what a frontier open weights model from OpenAI could look like. Seeing how @romainhuet and others take feedback and make decisions, it's no surprise they're moving so fast. Exciting times! x.com/sama/status/19…

Apr 11

people loved the GPT-4.5 podcast! what would you like to see one on next?

Apr 11

cute little robot

Apr 11

we have trained more models and they are good in some things

Quote

Jerry Tworek

@MillionInt

Sep 12, 2024

We trained a model and it is good in some things x.com/OpenAI/status/…

Apr 10

a lot of people were interested in how we made GPT-4.5 and what comes next. we did a podcast with alex paino, dan selsam, and

@atootoon

who helped drive the project. full episode coming soon, but here are some interesting clips:

Apr 10

here is the full video:

Sam Altman reposted

Apr 10

We’re releasing BrowseComp, which stands for Browsing Competition.

Think of it like coding or math competitions — while these contests may not perfectly reflect real-world SWE or mathematical research, they do capture a spark of intelligence. This is THE benchmark we should

Quote

OpenAI

@OpenAI

Apr 10

We’re open-sourcing BrowseComp (“Browsing Competition”), a new, challenging benchmark designed to test how well AI agents can browse the internet to find hard-to-locate information. It’s like an online scavenger hunt…but for browsing agents. openai.com/index/browseco

Apr 10

openai usage has gone nuts over the past month. but a surprisingly cool thing is you can ask the internet "hey can we please get hundreds of thousands of more GPUs quickly" and...the internet will deliver.

Apr 10

thank you all for the patience, and service speed and rate limits should continue to improve

Apr 10

we have greatly improved memory in chatgpt--it can now reference all your past conversations! this is a surprisingly great feature imo, and it points at something we are excited about: ai systems that get to know you over your life, and become extremely useful and personalized.

Apr 10

rolls out today for pro users and soon for plus users. (except users in the EEA, UK, switzerland, norway, iceland, and liechtenstein. call me, liechtenstein, let's work this out!)

Apr 10

you can of course opt out of this, or memory all together. and you can use temporary chat if you want to have a conversation that won't use or affect memory.

Apr 10

Apr 10

Apr 10

a few times a year i wake up early and can't fall back asleep because we are launching a new feature ive been so excited about for so long. today is one of those days!

Sam Altman reposted

Apr 9

we're launching the openai pioneers program – a partnership between openai and companies building advanced ai products to (a) intensively fine-tune models that outperform at high value domain-specific tasks, and (b) build better real world evals that enable industries to better

Quote

OpenAI Developers

@OpenAIDevs

Apr 9

Announcing @OpenAI Pioneers, a new program for ambitious companies building with our API. Selected teams will partner closely with us on domain-specific evals and custom fine-tuned models to advance AI product intelligence in their verticals. Apply to the first cohort below

Sam Altman reposted

Aleksander Madry

Apr 8

If AGI is about AI transforming our economy—how close are we, really? What's still missing, and how do we get there? OpenAI's new Strategic Deployment team tackles exactly these questions. We push frontier models to be more capable, reliable, and aligned—then deploy them to

Apr 7

we have shipped a lot of stuff recently, but people really love this in particular:

Quote

Atty Eleti

@athyuttamre

Mar 11

Introducing the Responses API: the new primitive of the OpenAI API. It is the culmination of 2 years of learnings designing the OpenAI API, and the foundation of our next chapter of building agents.

Here’s the story of how we designed it:

Apr 5

i am really loving spending more with research and product this is the most fun i've had since the early days when openai was just a quiet little research lab

You might like

Marc Andreessen

@pmarca

Click to Follow pmarca
Y Combinator

@ycombinator

Click to Follow ycombinator
Paul Graham

@paulg

Click to Follow paulg

Trending now

What’s happening

Sports · Trending

Slaf

Sports · Trending

The UFC

43.2K posts

Sports · Trending

Boring

63.7K posts

Combat Sports · Trending

Double Champ

5,038 posts